Imaging hydrogen peroxide in Alzheimer’s disease via cascade signal amplification

In brains of Alzheimer’s disease (AD), reactive oxygen species (ROS) levels are significantly higher than that of healthy brains. Evidence suggests that, during AD onset and progression, a vicious cycle revolves around amyloid beta (Aβ) production, aggregation, plaque formation, microglia/immunological responses, inflammation, and ROS production. In this cycle, ROS species play a central role, and H2O2 is one of the most important ROS species. In this report, we have designed a fluorescent imaging probe CRANAD-88, which is capable of cascade amplifying near infrared fluorescence (NIRF) signals at three levels upon interacting with H2O2 in AD brains. We demonstrated that the amplification was feasible in vitro and in vivo. Remarkably, we showed that, for the first time, it was feasible to monitor the changes of H2O2 concentrations in AD brains before and after treatment with an H2O2 scavenger. Our method opens new revenues to investigate H2O2 in AD brains and can be very instructive for drug development.

how the concentration of H 2 O 2 changes under disease conditions such as AD 20 . Numerous imaging probes have been reported for detecting H 2 O 2 in cells and in peripheral targets; however, none has been reported to detect H 2 O 2 in brains under different disease status [21][22][23][24][25][26][27] . Clearly, it is an urgent need to provide an in vivo imaging tool to detect the H 2 O 2 concentration in the living brain, particularly AD brains.
Several strategies for detecting H 2 O 2 have been developed, and Chang's boronate method and Nagano's diketone method are the most validated approaches [25][26][27] . Although the reported H 2 O 2 imaging probes are great for peripheral application and cell imaging [25][26][27] , the visualization of H 2 O 2 in the brain is not very straightforward. Therefore it is still challenging to use them for in vivo imaging of H 2 O 2 in AD brains.
In this report, we designed a curcumin analogue CRANAD-88, which has a capacity of three levels of cascade signal amplification. Remarkably, we demonstrated for the first time, that the changes of H 2 O 2 concentrations from AD brains could be monitored before and after H 2 O 2 scavenging. We believe that our method will be a valuable tool for investigating H 2 O 2 in AD brains, and for AD drug development and treatment regime design.

Results and Discussion
Design of fluorescence imaging probe CRANAD-88. Other researchers have reported various strategies to design H 2 O 2 responsive probes [25][26][27] . Among them, boronate oxidation is one of the most validated approaches for the design of imaging probes that can increase fluorescent intensity after reaction with H 2 O 2 25,26 . However, utilizing this method for cascade signal amplification has not been explored. In our previous studies, we demonstrated that CRANAD-58 was a "smart" NIRF probe for Aβ s, and displayed about a 100-fold intensity increase after binding to Aβ s [28][29][30][31] . To achieve cascade signal amplification, we integrated our previous Aβ "smart" probe strategy into the design of an H 2 O 2 responsive probe. Through the integration, CRANAD-88 was designed by linking a close mimic of CRANAD-58 (Aβ responsive moiety) to a boronate (H 2 O 2 responsive moiety) via a carbamate moiety (Fig. 1b). The benefit of the carbamate linker is that it leads the designed CRANAD-88 to be "invisible" in the near infrared spectrum. This invisibility is due to the carbamate's electron-withdrawing effect, which shortens the probe's excitation/emission wavelength. However, it becomes "visible" once it reacts with H 2 O 2 , because the boronate and carbamate moieties are removed to generate an electron-rich amino group for red-shifting. The probe's reaction with H 2 O 2 , and the subsequent binding of the reaction intermediate with Aβ s, as well as the red-shift all contribute to the cascade amplification of the NIRF signal (Fig. 1b). Fig. 2a, we first prepared the boronate-attached benzaldehyde via Curtius reaction with DPPA (Diphenylphosphoryl azide) in a high yield (67%). CRANAD-88 was obtained based on our previous published procedures [28][29][30][31] . The excitation spectrum showed that its excitation peak is around 580 nm and emission was about 690 nm in PBS buffer with 20% DMSO (SI Fig. 1).

Synthesis and spectral characterization. As shown in
In vitro solution testing for validating cascade fluorescence amplification. The emission peak of CRANAD-88 is around 690 nm, but its excitation peak is around 580 nm, therefore it is "invisible" for NIRF imaging (Ex > 640 nm required). To investigate whether CRANAD-88 is H 2 O 2 -responsive, we incubated CRANAD-88 in PBS buffer with various concentrations of H 2 O 2 . We found that the fluorescence intensity of CRANAD-88 was increased 3-fold after incubation with H 2 O 2 (100 μ M) at 37 °C (the first signal amplification) (Fig. 2b). We also observed significant red-shifts for both excitation and emission (Ex = 630 nm, Em = 730 nm) after the incubation (Fig. 2b). These red-shifts are very important for in vivo imaging, since a longer emission will allow imaging at greater depths and reduce background signal. Furthermore, these red-shifts convert "invisible" CRANAD-88 into a "visible" intermediate, which allows the probe to be visualized in vivo with NIRF imaging (the indirect signal amplification). To investigate whether the generated intermediate from the above incubation was responsive towards Aβ s, we incubated the above solution with Aβ 40 aggregates. As we expected, the fluorescence intensity of the intermediate was dramatically increased (2.5-fold) upon exposure to the aggregates (the second signal amplification) (Fig. 2b). To investigate whether the amplifications could be observed with parameters for in vivo imaging (Ex = 640 nm, Em = 700 nm), we conducted imaging (Ex = 640 nm, Em = 700 nm) with a plate that contained 1) CRANAD-88 only, 2) CRANAD-88 with H 2 O 2 , and 3) CRANAD-88 with H 2 O 2 and Aβ 40 aggregates. Consistent with solution spectral tests, significant increases could be observed from wells containing H 2 O 2 (4-fold), and H 2 O 2 with Aβ s (16-fold) (Fig. 2c,d). Remarkably, amplifications from the plate imaging were significantly higher than that from solution spectra at 700 nm, due to the "invisibility" of CRANAD-88 with the imaging parameters. Our in vitro data suggested that three levels of cascade fluorescence signal amplification were attainable through H 2 O 2 -reponsive interactions, Aβ -responsive interactions, and transformation of the probe from "invisible" to "visible" in the near infrared spectrum.
Consistent with other reported boronate imaging probes 25,26 , CRANAD-88 showed higher H 2 O 2 selectivity over other ROS species that include NO• , O 2 , ClO − (SI Fig. 2a). It is also very sensitive to H 2 O 2 , a significant increase could be seen with 10.0 μ M H 2 O 2 (SI Fig. 2b). In addition, CRANAD-88 exhibited quick responses towards H 2 O 2 , and a significant increase was observed only after 10 minutes of incubation (SI Fig. 2c).

Fluorescence responses of CRANAD-88 with Aβs without H 2 O 2 . When we incubated CRANAD-88
with various Aβ s, but without H 2 O 2, increases in fluorescence intensity were observed. However, we also observed significant blue-shifts to 640 nm (SI Fig. 3). The data indicates that, without H 2 O 2 , CRANAD-88 was "invisible" in the presence of Aβ s under conventional NIRF imaging parameters (Ex > 640 nm and Em > 650 nm required), due to its short excitation and emission.
Phantom imaging with mouse brain homogenate. To validate whether CRANAD-88 can be used in a biologically relevant environment, we used brain homogenates from a wild type mouse to investigate the changes of CRANAD-88 after exposure to H 2 O 2 , Aβ s, and both H 2 O 2 and Aβ s. We used an IVIS imaging system to collect data (Ex = 640 nm, Em = 700 nm). We observed an apparent signal increase after 30 minutes of incubation with H 2 O 2 (500 nM), but no significant signal increase was observed after incubation with Aβ s alone (Fig. 3a,b). This was because the probe was "invisible" with the imaging parameters until it was "turned on" by reaction with H 2 O 2 ( Fig. 3a,b). To confirm the cascade signal amplification in the homogenate, we further incubated the H 2 O 2 containing homogenate with Aβ 40 aggregates, and another significant increase of intensity could be observed (Fig. 3a,b). Taken together, the data suggested that the cascade signal amplification could be achieved in a biologically relevant environment.
In vitro mouse brain slice microscopic imaging. To investigate whether we can observe the fluorescence intensity changes of CRANAD-88 in AD mouse brain slices, we first incubated the probe with a brain slice of an 11-month old APP/PS1 mouse for 30 minutes. There was no significant intensity increase compared to the slice before probe treatment. However, we found that several plaques lit up once the same brain slice was incubated with H 2 O 2 at 37 °C for 30 minutes (Fig. 3c,d), and the lit-up plaques were well co-localized with thioflavin S staining (SI Fig. 4a,b). We quantified the signal using the fluorescence intensity ratio of the plaque of interest vs. an ROI of the same size of the vicinity of the plaque. We found that the amplification of the signal was about 2-fold with Aβ plaques and H 2 O 2 when compared with the background of pre-probe treatment or CRANAD-88 only (Fig. 3f). These results strongly indicated that signal amplification could be observed in an AD brain slice, and is also Aβ specific. BBB penetrating studies. To investigate whether CRANAD-88 can cross BBB, we injected the probe into a wild type mouse, which was sacrificed and perfused at 30 minutes after the injection. The dissected brain was homogenized and extracted with ethyl acetate. Fluorescent spectral testing and LC-MS data from the brain extraction confirmed that CRANAD-88 was able to cross the BBB (SI Fig. 4c-e).
In vivo imaging with CRANAD-88. Few references are available for the concentration of H 2 O 2 in the brain of living beings. It is not clear how the concentration of H 2 O 2 changes under disease conditions such as AD. Studies suggest that the concentrations of H 2 O 2 in AD brains are higher than that of normal brains [5][6][7] . However, direct evidence is still lacking. To prove the existence of elevated H 2 O 2 in the AD mouse brain, we used CRANAD-88 to conduct NIRF imaging with transgenic AD model APP/PS1 mice, a well-studied transgenic model for AD research 32,33 . Age-matched wild type mice were used as the control group. We monitored the NIRF signal at 15-, 30-, 60-, 120-, and 240-minutes post injection. We found that the APP/PS1 group showed significantly higher fluorescence signal at all of the time points (1.56-, 1.62-, 1.51-, 2.32-, and 10.91-folds) than WT mice. This data indicated that the difference between APP/PS1 and WT reflected by CRANAD-88 was likely due to the cascade NIRF signal amplification (Fig. 4a,b).

Monitoring H 2 O 2 scavenging in vivo.
To further confirm whether the above amplification is due to the existence of H 2 O 2 in the brain, we used sodium pyruvate, which is a widely used H 2 O 2 scavenger for cell culture 34 , and is capable of penetrating the BBB 35 . To conduct the experiments, we injected a mixture of CRANAD-88 and sodium pyruvate intravenously. As expected, the difference between the signals from the APP/PS1 group and the control group is significantly decreased (Fig. 4c,d, red bar), strongly indicating that amplification in vivo is due to H 2 O 2 triggering and engagement with Aβ s. In order to further investigate whether H 2 O 2 would be re-produced after the clearance of sodium pyruvate, we imaged the same groups of mice after one month. We found that the difference between APP/PS1 and WT was restored (Fig. 4d, green bar), suggesting that H 2 O 2 can be re-produced if toxic Aβ species are still existing.
Longitudinally monitoring the changes of H 2 O 2 concentration could be very important for investigating the effects of H 2 O 2 on the progression of AD pathology. From the design of CRANAD-88, it is clear that the total NIRF signal originates from the interaction with H 2 O 2 and Aβ s, and it is known that both Aβ and H 2 O 2 concentrations change with ageing; therefore, the NIRF signal changes from CRANAD-88 can't be used to quantitatively reflect the concentration change of H 2 O 2 in the longitudinal studies. To quantitatively calculate the relative H 2 O 2 changes during ageing and longitudinal therapy monitoring, we will use the ratio between CRANAD-88 and CRANAD-3 (R H 2 O 2 = R (CRANAD-88) /R (CRANAD-3) ). For CRANAD-3, our previous data showed that it could reflect the changes in Aβ s during ageing and treatment 31 . Using the ratio of the two probes, we will be able to reliably calculate relative H 2 O 2 changes in the longitudinal studies. Summary H 2 O 2 plays important roles for AD pathology. However, no method is available to detect the changes of H 2 O 2 in the brain. In this report, results from in vitro spectral studies, plate imaging, phantom imaging, and brain slice imaging consistently suggested that three levels of cascade signal amplification was feasible. Remarkably, for the first time, we demonstrated that the changes of H 2 O 2 concentrations in AD brain could be monitored with a NIRF imaging probe capable of cascade signal amplification. Our study is not only crucial for the proof of the existence of H 2 O 2 , but also points out several important questions that are needed to be addressed in terms of seeking effective treatments for AD. Our study strongly indicated that H 2 O 2 can be reproduced if the Aβ plaques or other Aβ s are still existing, and this pointed out that AD treatment is not only related to the removal of Aβ s, but also related to ROS. Our study also pointed out another important question: whether Aβ -lowering treatment can lead to the decrease of ROS such as H 2 O 2 . If the Aβ -removal cannot reduce the ROS level, the toxic ROS will continuously damage the neurons. Therefore, ROS reduction should be considered to be one type of treatment outcome to be evaluated. In summary, our method can be used to monitor the changes of H 2 O 2 in brains of AD mice, and this technology will be an important tool for AD drug development and treatment regime management.

Materials and Methods
Reagents used for the synthesis were purchased from Sigma-Aldrich and used without further purification. The pH of the PBS buffer was 7.4. Column chromatography was performed on silica gel (SiliCycle Inc., 60 Å, 40-63 mm) slurry packed into glass columns. Synthetic Aβ peptides (1-40/42) were purchased from rPeptide (Bogart, GA, 30622). Aggregates for in vitro studies were generated by the slow stirring of Aβ 40 in PBS buffer for 3 days at room temperature. CRANAD-88 was dissolved in DMSO to prepare a 25.0 μ M stock solution. 1 H and 13 C NMR spectra were recorded at 300 and 125 MHz, respectively, and reported in ppm downfield from tetramethylsilane. Fluorescence measurements were carried out using an F-4500 fluorescence spectrophotometer (Hitachi). Transgenic female APP-PS1 mice and age matched wild-type female mice were purchased from (c) Representative images of WT and APP mice, which both were treated with sodium pyruvate, d) the NIRF signal differences of CRANAD-88 between AD and WT mice (F (APP) − F (WT) ) treated without (blue bar) and with sodium pyruvate, and re-appearance of H 2 O 2 one month after pyruvate scavenging (green bar). P value: *< 0.05, **< 0.01, and ***< 0.005. Jackson Laboratory. All animal experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at Massachusetts General Hospital, and carried out in accordance with the approved guidelines.
Step 1: 1.0 mL of PBS buffer with 20% DMSO was added to a quartz cuvette as a blank control and its fluorescence was recorded with the same parameters as for CRANAD-88.
Step 2: The fluorescence emission spectrum of a CRANAD-88 solution (1.0 mL, 250 nM) was recorded with excitation at 560 nm and emission from 580 to 900 nm.
Step 3: To the above incubated CRANAD-88 solution, 10 μ L H 2 O 2 was added (final concentration 100 μ M). The emission spectra at different time points were recorded (Ex = 560 nm, Em = 580-900 nm). Step 4: To the above solution of CRANAD-88 and H 2 O 2 , 10 μ L of Aβ 40 aggregates were added to make the final Aβ concentration of 250 nM. The emission spectra of the resulting mixture at different time points were recorded as described in Step 2. The final spectra from steps 2 and 3 were corrected using the blank control from Step 1.
BBB penetration of CRANAD-88. Similar procedure from ref. 30 was followed. The ethyl acetate extraction was subjected to fluorescence spectra recording and LC-MS analysis.
Brain phantom studies. A 4-month old balb/c mouse was perfused and sacrificed, and its brain was dissected. The brain was homogenized with 2.0 ml PBS, and the homogenate was equally divided into five eppendorf tubes and diluted to 1 ml with PBS. Every tube had added 20 μ L CRANAD-88 (final concentration 500 nM) except the control of homogenate blank. No Brain slice testing. A brain slice (25 micron) of an 11-month old APP/PS1 mouse was washed with double-distilled water for 5 minutes, then fixed with 4% formalin for 5 minutes, and washed with double-distilled water. First, a background image of a brain slice was taken using a fluorescence microscope before probe treatment. Second, an image was taken after the slice was incubated with CRANAD-88 (25 μ M) for 30 minutes. Third, an image was taken after the above slice was incubated with H 2 O 2 (100 μ M) at 37 °C for 30 minutes.
In Vivo NIRF Imaging. In vivo NIRF imaging was performed using an IVIS Spectrum animal imaging system (Caliper LifeSciences, Perkin-Elmer, Hopkinton, MA). Images were acquired with a 640 nm excitation filter and a 700 nm emission filter. Data analysis was performed using Living Image 4.2.1 software. 15-month old transgenic APP-PS1 mice (female, n = 3-4) and age-matched wild-type control mice (female, n = 3-4) were shaved to remove the fur from the brain area before background imaging. An injection solution of CRANAD-88 (4.0 mg/kg) was freshly prepared in 15% DMSO, 15% cremorphor, and 70% PBS, and the solution was stabilized for 20 min before injection. Each mouse was intravenously injected with 100 μ L of CRANAD-88 via tail vein. Fluorescence signals from the brain areas were recorded at 0-, 15-, 30-, 60-, 120-, and 240-min after the intravenous injection. To quantify the NIRF signal, an equal size ROI was drawn around each brain region.
Monitoring H 2 O 2 scavenging in vivo. The same mice from the above experiments were used after waiting for one week. In vivo imaging data indicated that the injected CRANAD-88 was totally washed out. For scavenging